How We Rate Crypto Exchange Providers on Swapzone

Summarize with AI

Methodology version 1.0

Every comparison site shows you providers with 4.6 stars and others with 4.3, yet no one explains the differences between them.
Pick by the star, and you may still wait 50 minutes for a swap and receive 2% less than quoted, because the number reflected marketing, affiliate payouts, or a reviewer’s afternoon, not what happens to your funds. This page explains how Swapzone rates crypto exchange providers, detailing what each of the five numbers on a provider page measures, where the data originates, how the total score is calculated, and which factors cannot influence it.

Key facts

• Crypto exchange provider ratings on Swapzone use five metrics, each scored 1.0–5.0: transaction speed, equal to estimated, refund policy, partner support, and reviews.
• Total score is the plain average of the five. No hidden weights.
• Data comes from swaps completed through Swapzone, refund cases our support team handled, escalations to the provider, and reviews tied to a transaction.
• We recalculate scores monthly using a rolling 90-day window.
• Affiliate commissions are identical for every provider and cannot change a score. Providers cannot pay for a rating.

Contents

1. What a score on a provider page means
2. Where the data behind crypto exchange provider ratings comes from
3. The five metrics
4. How the total score is calculated
5. What cannot change a score
6. What we show but do not score
7. How often scores change and how to dispute one
8. Why the method looks like this
9. FAQ

What a crypto exchange provider rating on Swapzone means

A crypto exchange provider rating on Swapzone answers one question: what happens to your funds after you click “exchange” with this provider. It does not measure how many coins a provider lists, how its app looks, or how loudly it advertises. Those things are visible before you send money. Speed, accuracy, refunds, and support are not, so those are what we score.

All scores derive from events that occurred through Swapzone.
Since 2019 we have routed swaps to 18+ providers across 1,600+ cryptocurrencies without holding funds or executing trades ourselves. That position gives us something a review site cannot get: the timestamps, amounts, and outcomes of real swaps, plus every refund and support case our team escalated to a provider.

The label “Verified rating · based on N swaps via Swapzone” on a provider page, ChangeNOW’s page for example, means exactly that. Where a provider has too little history, the page says “Not enough data” for that metric and “Provisional” next to the total, rather than showing a default number that looks like an assessment.

Where the data behind crypto exchange provider ratings comes from

Three sources are deliberately absent. Provider self-descriptions (“swaps in 5 minutes”, “24/7 support”) are not data. Third-party review sites are not used for partner scores because their reviews are not tied to a swap and can be gamed. Our own opinion of a provider is not a source either: the account manager who talks to providers daily can flag a case, but the case must be in the log with an ID before it changes a number.

The five metrics

Transaction speed

The swap record measures speed, rather than the provider’s claim.
Every swap completed through Swapzone carries the time it actually took from creation to payout, and that record is the input. The time you see next to an offer before you swap is the same data in a shorter window: the provider’s average on that pair over the last 30 days.

Compared pair by pair

Providers are compared pair by pair, not on one global average. A provider that handles many Monero swaps would look slow next to one that mostly sends USDT on Tron, and that would tell you nothing. Comparing within a pair also removes the part of the clock the provider does not control: deposit confirmation on the sending network takes the same time whichever provider is on the other end. For every pair with enough volume, we compute the market median time across all providers, then the provider’s own median on the same pair. The ratio, weighted by the provider’s swap count, is its speed index. An index of 1.0 means “as fast as the market on the same routes”.

Penalty for stuck swaps

The tail matters more than the median, so it carries a separate penalty. Swaps that take more than six times the market median for their pair count as stuck. If more than 1% of a provider’s swaps are stuck, the score drops by 0.3; above 3% by 0.6; above 5% by 1.0. Bitcoin.com Research × Swapzone data on 150,000 swaps shows execution speed varying up to 45x between providers and an industry median of 45 minutes, which is why a provider’s average alone would hide the swaps that cost users the most.

Swaps that ended in a refund or were held for compliance checks are excluded here and counted under Refund policy. The metric needs at least 30 completed swaps in 90 days; below that, the page shows “Not enough data”.

Equal to estimated

This metric answers whether you receive what the offer showed. Every swap record holds two pairs of numbers: the amount the user was asked to send and actually sent, and the amount the offer promised and the provider actually paid out. The second pair is the metric. A swap counts as a hit if the received amount is within 0.5% of the estimate or above it.

Rare but large misses are penalised separately. If the average shortfall on missed swaps exceeds 3%, the score drops by 0.5; above 5%, by 1.0. A provider that misses one swap in twenty by 8% is worse for you than one that misses one in ten by 0.7%, and a hit rate alone would rank them the other way round.

What is excluded

Two kinds of swaps are excluded so that the provider is not blamed for the user or the market. Swaps where the deposit arrived more than 30 minutes after the offer was created are out, because a floating rate moves in that time regardless of the provider. Swaps where the first pair of numbers does not match, meaning the user sent more or less than entered, are out too; the payout could not match the offer, and such cases are handled as refunds. Fixed-rate swaps stay in and count as hits when executed at the fixed amount.

Refund policy

We score what happens when a swap fails, not what the provider’s terms page says about it. Every refund case that reaches Swapzone support is logged with the provider, the reason, the time it was opened, and the time funds were back with the user. That log is the only input.

Four reasons, two clocks

Reasons fall into four categories, because the fair deadline depends on the reason. A duplicate deposit can be returned in an hour. Funds sent on the wrong network have to be recovered manually. A compliance hold needs documents from the user before the provider can act. So each case gets points on one of two clocks:

From cases to a score

The provider’s score is the average of its case points over 90 days, pulled toward the all-provider average when the provider has few cases. A single slow refund among a handful of cases moves the score by tenths, not by whole points; ten slow refunds move it all the way. A separate deduction applies when refunds are the provider’s own fault, such as a liquidity partner failing to execute: up to 1.0 off, based on how many such cases occur per 1,000 completed swaps.

Why the written policy is not the metric

One case from our support log shows why the written policy is not the metric. A user sent a swap as two separate transactions instead of one, which most terms pages treat as unrecoverable. Swapzone support contacted the provider, and the user was offered a full refund at the original rate or continuation at the current rate. The funds came back. That outcome, and its timing, is what the score records. The provider’s terms page would have predicted a loss.

Partner support

This metric measures how a provider responds when Swapzone escalates a user’s problem to it. It is not a score of the provider’s own customer service, which we cannot observe. It is a score of the channel that decides whether your stuck swap gets fixed today or next week.

The base score comes from a rubric filled in quarterly by our account management team, with every criterion tied to a record:

Monthly adjustments are logged with a case ID. A critical case left unanswered for more than three business days costs 0.3, capped at 1.0 per month. An exceptional resolution of a hard case adds 0.1, capped at 0.2. Adjustments expire after one month, so a bad week does not follow a provider for a year.

Reviews

Only reviews tied to a completed transaction count. A user can leave one review per swap, the review is linked to the transaction ID, and reviews flagged as spam by moderation are excluded. Provider replies to reviews do not affect the score.

The score is the average rating over the last 12 months, pulled toward the all-provider average when the review count is low. Twenty reviews is the point where a provider’s own average starts to dominate. Below ten reviews the page shows “Not enough data”. This is why a provider with three five-star reviews does not outrank one with 260 reviews averaging 4.7.

How the total score is calculated

The total crypto exchange provider rating is the average of the five metrics, each weighted equally. We publish no hidden weights because we have no evidence yet that speed matters more to users than refunds or the other way round, and because unequal weights are the easiest place to tilt a rating without anyone noticing. If data from a year of scoring supports different weights, the change will appear in the changelog on this page with the reasoning.

Rolling 90-day window

Every metric runs on a rolling window and is recalculated on the first working day of each month. Transaction metrics and refunds use the last 90 days; reviews use 12 months. A month is too short: a provider can have one refund case in July and a dozen in August. A quarter is long enough to see a pattern and short enough that an improvement shows within a season.

Three rules for thin data

Three rules keep thin data from producing confident-looking numbers. A metric below its minimum sample shows “Not enough data” and is left out of the average. A provider with fewer than three scored metrics or less than 90 days of history carries a “Provisional” label next to its total. And metrics built from small counts, refunds and reviews, are smoothed toward the all-provider average, so one event changes a score by tenths and a trend changes it by points.

The “30d” figure under the total is the change against the previous month’s total. Monthly history is kept so that a provider’s page can show whether it is improving.

What cannot change a crypto exchange provider rating

Money cannot. Swapzone earns a fixed affiliate commission from each provider at the same rate regardless of which provider a user picks, so there is no revenue reason to prefer one score over another. Providers cannot buy a rating, a rank, or a badge, and there are no sponsored placements on provider pages.

A provider’s request cannot either. Providers can dispute a score, and we want them to, but a dispute is resolved with data: specific swap IDs, refund cases, or escalation records that we then re-check. “Our team feels the score is too low” is not an input. Corrections are made with a date and a changelog entry, never as a silent edit.

Three hard limits sit above the formula, because some failures should not be averaged away. Two or more confirmed cases in 90 days of funds not returned without a legal basis cap Refund policy at 2.0 and the total at 3.5. A security incident with user losses caps the total at 3.0 until the provider publishes a post-mortem and compensates users. Three or more escalations left unanswered for more than five business days in 90 days cap Partner support at 2.0. This follows the principle used in institutional exchange benchmarks such as the CoinDesk Data Exchange Benchmark: weakness in a critical category limits the overall grade and cannot be offset by strength elsewhere.

What we show but do not score

KYC and AML practices appear as labels on every provider page and never enter the score. Whether a provider requires identity verification is a fact you need before you send funds, but it is not a quality. A privacy-focused user and a user in a regulated jurisdiction read the same label and make opposite choices, and both are right. Scoring it would mean deciding for you.

The labels follow a fixed vocabulary so that providers can be compared at a glance. KYC: Not required, Not required up to a stated amount, Required, Required for large amounts, Varies by country. AML: Disclosed, Licensed, Not disclosed. Labels are set from the provider’s confirmed policy and reviewed quarterly. This mirrors how CoinGecko’s Trust Score methodology treats attributes it tracks but does not score.

How often scores change and how to dispute one

Scores update monthly, and the “Last reviewed” date on each provider page is the date of the last recalculation, not the date someone edited the text. An out-of-cycle recalculation happens within 48 hours after a security incident, a halt in payouts or refunds, regulatory action against a provider, or a provider leaving Swapzone.

Providers can request the breakdown behind their score: metric values, the window, and the number of swaps or cases involved. Disputes go to [email protected] with the transaction or case IDs in question, and we answer within five business days. We list any corrections in the changelog below.

Swapzone versions the rating methodology itself.
Any change to a threshold, a weight, a data source, or a rule creates a new version with an entry stating what changed and why. Earlier versions are available for users to review a provider’s score history against the rules that produced it.

Why the method looks like this

The metrics follow the money users actually lose. In the Bitcoin.com Research × Swapzone study of 150,000 swaps, users on the slowest providers lost an average of 3.2% of the swap value to rate movement during execution, against 0.1% on the fastest. That gap is larger than most providers’ entire fee, which is why speed and estimate accuracy are two of the five metrics rather than footnotes.

The rules for crypto exchange provider ratings borrow from benchmarks that publish their scoring instead of asserting it. The idea that critical failures cap the total rather than being averaged comes from the CoinDesk Data Exchange Benchmark. Tracking a property without scoring it, as we do with KYC, is how CoinGecko handles attributes such as team presence. Equal weights, versioning, and public thresholds are the parts of editorial rating methodologies at Investopedia and NerdWallet that make them checkable, and we adopted them for the same reason.

What we did not borrow is the input. Those benchmarks rate exchanges from public disclosures, questionnaires, and order-book data. A swap aggregator sees the swaps themselves. Building the rating on that record, and publishing the rules, is the only way a score can mean something when the provider it describes also pays us a commission.

FAQ